International Journal of Rock Mechanics & Mining Sciences 67 (2014) 164–169

Contents lists available at ScienceDirect

International Journal of
Rock Mechanics & Mining Sciences
journal homepage: www.elsevier.com/locate/ijrmms

Experimental study of gravity flow under confined conditions

Raúl L. Castro a,n, Miguel A. Fuenzalida a, Fernando Lund b
a
Block Caving Laboratory, Advanced Mining Technology Center, University of Chile, Santiago, Chile
b
Physics Department, Faculty of Physical Sciences and Mathematics, University of Chile, Santiago, Chile

ar t ic l e i nf o a b s t r a c t

Article history: Caving methods rely on gravity to break up and transport large amounts of ore and waste. Massive
Received 26 March 2013 underground mining is becoming even more prevalent due to the depletion of surface mining reserves.
Received in revised form It is thus relevant to study the gravity flow mechanisms that will occur at deeper levels. Despite the
10 January 2014
importance of gravity flow, there is a lack of quantification of the influence that confinement (the weight
Accepted 20 January 2014
of the ore column) has on the secondary fragmentation and the caved rock ability to flow. This paper
Available online 7 March 2014
presents the design and results of an experimental setup used to investigate the flow mechanisms of
Keywords: cohesionless material when drawing from a single drawpoint under confinement. Experimental results
Gravity flow mechanics showed that the flowability of the material is influenced by the rock fragment size, dimension or
Isolated draw
diameter of the opening and the vertical load applied. Secondary fragmentation is mainly influenced by
Interlocking arching
the vertical load applied and the size distribution of the fragmented rock. Finally, flowability mechanisms
Hang-ups
Secondary fragmentation are presented in terms of a state graph and a hang-up frequency graph, both of which could be used for
the design of openings in mining.
& 2014 Elsevier Ltd. All rights reserved.

1. Introduction of a number of hang-ups per 1000 t drawn and in terms of average

extracted tonnage in between hang-ups [7].
Block caving is an underground bulk mining method that offers The process of fragmentation in a cave mine involves both
continued production at a comparably high rate, with low operat- primary and secondary fragmentations. Primary fragmentation is
ing costs. Block caving relies on gravity to break up and transport attributable to the induced stresses during caving propagation
large amounts of ore. In order to achieve a continued high rate of while secondary fragmentation results from the breakage process
production, there are certain key aspects that need to be taken into that occurs as the blocks move down through the draw column to
account in the design of a caving operation. One of these is the the drawpoints [1].
fragmentation that can result from the induced stresses during In terms of the mechanisms of secondary breakage, Pierce [8]
caving propagation and the associated movement of the caved has postulated that there are three main modes of breakage in a
rock to the drawpoints. Another, related aspect, is the flow cave: impact breakage within the air gap at the top of the cave;
capability of the caved rock [1,2]. compression breakage within stagnant zones of the cave resulting
Flowability can be defined as the flow condition or ability of a in block splitting (a fragment breaks into two or more pieces); and
granular material to flow under a given set of material properties, shear-driven breakage within movement zones resulting in corner
infrastructure geometry and stress conditions. In the past, flowability rounding. The relative influence of each kind of breakage remains
has been classified into free flow, intermittent flow, and no-flow. uncertain.
The methods for estimating flowability have been derived from To date, knowledge on the mechanisms of gravity flow in caving
observations at mine sites, and experiments have shown particle mines (flowability and secondary fragmentation) has been gained
size, opening size and fines content to be key parameters. The through mine data analysis, numerical modeling and physical model-
flowability, in turn, defines the size of a key mine infrastructure, the ing. Physical models have the advantage to be controlled experiments
diameter of ore passes or drawpoints [3–6]. In the case of intermittent that can be used to understand the mechanisms of flow. Significant
flow it is useful for mine operations to have a quantification of efforts have been made to build physical models to understand the
interferences or hang ups, that determine the intermittency, in terms mechanisms of gravity flow in caving mines including large-scale 3D
models [9,10]. However, even the large physical models studied to
date do not achieve full similitude to a real operation and cannot
be used to understand breakage and compaction since the vertical
n
Corresponding author. Tel.: þ 56 2 2978 4497. loads that have been used are relatively small (about 32 kPa) when
E-mail address: rcastro@ing.uchile.cl (R.L. Castro). compared to the strength of the granular media (100–150 MPa).

http://dx.doi.org/10.1016/j.ijrmms.2014.01.013
1365-1609 & 2014 Elsevier Ltd. All rights reserved.
 R.L. Castro et al. / International Journal of Rock Mechanics & Mining Sciences 67 (2014) 164–169 165

Under those conditions, hang-ups as a result of mechanical arches,

breakage and compaction are not observed. This article addresses this
shortcoming of previous experiments: by the use of confined flow
experiments – whose set up and initial results are presented below –
we aim at the understanding of these important mechanisms.

2. Laboratory equipment for confined granular flow

2.1. Equipment design

One of the main objectives of this research was to design a set

of experiments where flow tests could be performed under large
vertical stresses [11]. The confined flow set up consists of a steel
cylinder, an extraction system and a hydraulic press (Fig. 1). The
steel cylinder has 340 mm inner diameter, and can stand an
applied pressure of 14 MPa. The application of the load is through
a 1800 KN hydraulic press. The cylindrical geometry of the steel
cylinder was chosen to avoid boundary effects related to arching,
and stress concentrations at corners. In terms of arching the
diameter was chosen so that it could handle up to 12 mm particles.
Consequently the width to particle size ratio is 28, which is well
above the soil mechanics standards for triaxial tests of 10 particle
sizes [12]. In terms of the flow characteristics, the diameter of the
flow zone was always smaller than 340 mm so it was an isolated
draw. The height of the cylinder was 700 mm in order to hold the
desired volume of gravel and to comply with the 2:1 height to
width ratio. The base of the cylinder had a drawpoint located in
the center of the model so that the flow zones would not intersect
the model's walls. The diameter of the drawpoint (dw) was set at
100 mm. A steel extraction system was built to extract the material
(Fig. 1b), containing a wedge that pushes the material out of the
cylinder.

2.2. Preparation of specimens and procedure

The experiment required the application of an axial load,

ranging from 0 to 1800 KN, which is the maximum load capacity
of the hydraulic press. The piston is brought to rest against the
surface of the sample. The press is started and when it reaches the
required load, the extraction process begins and the load is kept
constant. One of the questions to be answered before the tests
were run was the magnitude of the vertical load that ought to be
applied to our press in order to mimic the vertical stresses that
occur when a large fragmented pile of rock develops after the cave.
In order to quantify the vertical stress in a granular material we
use Janssen's formula [13]
Rh γ
sv ¼ ½1  ekð tan ϕÞz=Rh
; ð1Þ
k tan ϕ
where sv is the vertical stress, Rh is hydraulic radius (area/
perimeter) of the area of draw under analysis, γ and ϕ are the
density and friction angle of the fragmented rock, z is the height of Fig. 1. Hydraulic press and steel cylinder (a) and the cylinder (b).
the caved rock pile, and k is an earth pressure related constant
whose value depend on the mode of failure of the granular
material. Taking Rh ¼70 m; γ ¼2200 kg/m3 and ϕ¼381, k ¼1- Table 1
sinϕ¼ 0.38, we tabulate the resulting stress together with the Range of loads according to vertical stresses using Eq. (1).
equivalent press load for the geometry of our apparatus in Table 1.
We conclude that our experiment represents the conditions of a Depth Mean vertical stress (MPa) Press load (KN)
cave pile of more than 1000 m height, which is considered deep.
100 1.76 159
The procedure consisted of applying a constant axial load 200 2.91 264
during the test until a determined quantity of material was 300 3.66 332
extracted from the drawpoint. Once the model media was tested, 500 4.47 406
the cylinder was removed from the press and placed in a 700 4.82 437
1000 5.00 454
horizontal position. Then it was opened and “cut in half” in order 1500 5.07 460
to observe flow zones and to take samples for fragmentation
166 R.L. Castro et al. / International Journal of Rock Mechanics & Mining Sciences 67 (2014) 164–169

Fig. 2. (a) Flow zone observed at the end of experiment, d50 ¼ 12 mm. (b) Flow zone observed at the end of experiment, 13.2 MPa.

analysis. Due to the 3D configuration of the physical model, the Table 2

material was surrounded on all sides and could not be visually Summary of the Characteristics of the Model Media used in the Experiments.
inspected from the outside. Therefore, in order to determine the
Media Mean size, Uniformity Swell, Drawpoint/particle
extraction zone, painted numbered tracers were positioned inside d50 index, Cu n [%] size ratio, w/d50
the model and recovered at the drawpoint (Fig. 2). In order to
determine the movement zone (flow zone), painted markers with 6 mm 6.25 1.5 44 16
different colors were placed every 10 cm (Fig. 2). 8 mm 8.17 4.1 40 12
12 mm 12.59 1.17 44 8
Regarding the flowability measurements for these tests, four
modes of flow were recorded. (a) Free Flow: the material flows
freely through the point of extraction without interruption. (b)
Intermittent Flow: the material flows with intermittent interrup-
tions due to arches formed by mechanical material compaction.
They break apart due to the vertical force applied by the hydraulic
press. (c) Assisted Flow: in order for the material to flow, manual
intervention is needed to disrupt the mechanical arches. (d) No
Flow: the material is completely stagnant, even when disturbed
manually.

2.3. Model media

The tests described in this paper were conducted using crushed

gravel. During the experimental phase, three model media were
tested, two with a fairly narrow distribution of different particle
sizes (d50 ¼6 mm and 12 mm) and a third with a wide distribution
(d50 ¼8 mm), as indicated in Table 2. The ratios between particle Fig. 3. Particle size distribution for the model media.
size and the opening diameter were 6, 8 and 12 respectively. The
cumulative size distribution curves for the different particle sizes
are presented in Fig. 3. porosity is the amount of voids in the granular media before
Other properties of the media under study were measured and the tests.
are summarized in Table 2. In this the uniformity index is the ratio The shear strength characteristics of the tested materials were
between the sixty and ten percent particle passing size, and the determined using a large (300 mm  300 mm) shear box. Samples
 R.L. Castro et al. / International Journal of Rock Mechanics & Mining Sciences 67 (2014) 164–169 167

were subjected to different normal forces ranging from 50 to results in a more difficult flow condition. Reduction was quantified
980 KPa. It was observed that cohesion was zero and the friction for the mean size in terms of the initial (d50,i) and final size (d50,f)
angle reached a value of 421. It was also found that for normal as
stresses above 480 KPa the gravel showed a low angle of dilatancy, d50;f d50;i
indicating that above this value the material would break due to R50 ¼ ð2Þ
d50;i
compaction. Other mechanical properties for the model media are
shown in Table 3. The size distributions of particles before and after the tests are
presented in Figs. 4–6. As expected, the greater the vertical load
applied, the greater the degree of particle breakage. Regarding the
3. Experimental results influence of the size distribution on the breakage, the results show
that breakage occurs mainly for smaller particles generating a
A total of twenty-three experiments were performed to quan- considerable amount of fines. However in the wide distribution,
tify confined flow, with results that are presented in Table 4. Each breakage occurs among all particle sizes reducing the overall mean
one of the three batches of gravel (6, 8, and 12 mm) was loaded to size.
2, 3, 5.8 and 12 MPa, twice, with the exception of the 8 mm batch
that was loaded to 5.8 MPa only once. After each loading material
was extracted from the drawpoint, as explained in Section 2, and 4. Flowability assessment of caved rock
its size distribution was determined (Table 4, columns 3, 4 and 5).
Also, the flow condition was noted in each case (Table 4, column From the experiments and data collected from the literature a
6). As expected, there is a general trend that larger particles flow flowability graph was designed as shown in Fig. 7. In this the
with mode difficulty. Importantly, the data shows that as the different flow states of the caved rock is a function of vertical
vertical load increases the mean size of the extracted material
decreases, and its dispersion increases. An increased load also

Table 3
Summary of the characteristics of the model media used in the experiments.

Gravel

Point strength Index I s50 (ASTM D5731-08) [MPa] 6.3 (7 1.6)

Particle shape Sphericity 0,65

Roundness 0,51
Regularity 0.58

Solid density [ton/m3] 2.69

Residual friction angle [deg.] 42

Peak friction angle [deg.] 48

Wall friction angle [deg.] 20

Fig. 4. Size distribution for 6 mm media.

Table 4
Summary of experimental results.

Exp. rv [MPa] d50i [mm] d50f [mm] R50[%] Flow condition Interferences [g/hang-up] Standard dev. [g/hang-up]

1.1.A 2 6.25 6.1 2.4 Free 955 156

1.2.A 3 6.25 5.93 5.1 Intermittent 625 0
1.3.A 5.8 6.25 5.7 8.8 Intermittent 703 559
1.4.A 12 6.25 5.29 15.3 Assisted 489 105
1.1.B 2 6.25 5.95 4.8 Free INF –
1.2.B 3 6.25 6.07 2.9 Free INF –
1.3.B 5,8 6.25 5.95 4.8 Intermittent 680 160
1.4.B 12 6.25 5.29 13.5 Assisted 546 127

2.1.A 2 8.17 7.07 13.5 Intermittent ND ND

2.2.A 3 8.17 7.06 13.6 Intermittent ND ND
2.3.A 5.8 8.17 6.77 17.1 Assisted ND ND
2.4.A 12 8.17 6.4 21.7 Assisted ND ND
2.1.B 2 8.17 7.82 4.3 Intermittent ND ND
2.2.B 3 8.17 7.71 5.6 Intermittent ND ND
2.4.B 12 8.17 6.51 20.3 Assisted ND ND

3.1.A 2 12.59 12.48 0.9 Intermittent 778 133

3.2.A 3 12.59 12.31 2.2 Assisted 383 225
3.3.A 5,8 12.59 12.31 2.2 Assisted 403 229
3.4.A 12 12.59 11.83 6.1 No flow 0 0
3.1.B 2 12.59 12.4 1.5 Free INF –
3.2.B 3 12.59 12.23 2.9 Assisted 493 445
3.3.B 5.8 12.59 11.85 5.8 Assisted 303 99
3.4.B 12 12.59 11.15 11.4 No flow 0 –

ND: no data was measured; INF hang ups were not observed.
168 R.L. Castro et al. / International Journal of Rock Mechanics & Mining Sciences 67 (2014) 164–169

stress and dw/d50 ratios. This was built using the above experi- particle size (dw/d50) is equal to 5. In this case we would expect to
ments and also the results published by Hustrulid [3] on ore pass have intermittent-flow with cero vertical load. As the vertical load
designs, which were conducted under near zero vertical loads. The increases (see Eq. (1) for reasons for this) the flow would change to
points for nonzero vertical load are extracted from Table 4. assisted flow at about 2 MPa while at 6 MPa would be in a no flow
Fig. 7 could have applications to cave mining. For example let condition. In this case the flow condition could be changed from
us consider that due to design, the ratio between drawpoint to no-flow to intermittent-flow by increasing the drawpoint so that
dw/d50 46.
In those of our experiments that presented intermittent flow,
hangups occurred depending on the vertical load. The influence of
stresses on the hang up frequency was quantified measuring the
mean amount of fragmented rock drawn before an arch occurs.
As noted in Fig. 8, the amount of mass drawn before a hang up

Fig. 5. Size distribution for 8 mm media.

Fig. 8. Hang-up frequency as a function of vertical load.

Table 5
Fitted parameters for hang up frequency.

Media dw/d50 Hgo (gr/hang up) r (adim)

6 mm 12.6 16.498 0.02

12 mm 6.25 111.125 1.44
Fig. 6. Size distribution for 12 mm media.

Fig. 7. Flow-ability graph of caved rock.

 R.L. Castro et al. / International Journal of Rock Mechanics & Mining Sciences 67 (2014) 164–169 169

occurs decreases as the vertical stress increases in a non-linear standard for mine design applications when quantifying confined
fashion. Also, as was expected, as the dw/d50 ratio increases there flow. To this end, the influence of scale and diameter of the
is a reduction in the hang-up frequency. Fig. 8 shows a fitted cylinder should be considered. Large physical models could still
exponential equation represented as follows: be used to understand the kinematic of flow including fines
H g ¼ H go expð  r sv Þ ð3Þ migration and production rates. These topics are currently under
research and will be submitted for publication in the near future.
where Hg is the amount of material that was drawn before a hang
up occurs, Hgo and r are fitted parameters which are related to the
hang up frequency of the fragmented rock under zero vertical
Acknowledgments
stress and the rate at which the flowability reduces with the load
respectively. Thus a larger value of r would mean larger decrease
on the flowability with the increase on vertical stress. Table 5 This paper describes a component of the work carried out
shows the value of the fitted parameters for the media analyzed in under the Project “Engineering Fundamentals of Block Caving”
these experiments. run by the University of Chile's Advanced Mining Technology
Center (AMTC) and funded by the Chilean Government through
Conicyt. The authors would also like to acknowledge Mr. Rene
5. Conclusions and discussion Gomez for helping in carry out many of the tests published in this
article.
The applied vertical stress was found to have a considerable
effect on the flowability and fragmentation of the media using
a set up that simulates the vertical loads that could occur in References
caving mines. Changes include reduction in size and flowability of
fragmented rock for increased loading. The flowability has been [1] Brown ET. Block caving geomechanics. 2nd ed. Brisbane: Julius Kruttschnitt
Minerals Research Center, University of Queensland; 2007.
measured using experiments and a state graph has been derived.
[2] Chitombo GP. Cave mining – 16 years after Laubscher's 1994 paper “Cave
From this graph, the flowability of a given fragmented material mining – State of the Art”. In: Potvin Y, editor. Caving 2010. Australia: Perth;
could be classified for a given vertical stress and particle/opening 2010. p. 45–61.
size ratio. [3] Hustrulid W, Sun C. Some remarks on ore pass design guidelines. In:
Karzulovic A, Alfaro M, editors, Proceedings of MassMin, Santiago, Chile;
Flowability was quantified for the case of intermittent flow in 2004. p. 301–8.
terms of hang up frequency. The hang up frequency depends on [4] Kvapil R. Gravity flow of granular materials in hoppers and bins. Int J Rock
the particle to opening size ratio, and on the vertical stress. Mech Sci 1965;35:41–2.
[5] Hadjigeorgiou J, Lessard JF. Numerical investigations of ore pass hang-up
The experimental results indicate that large vertical stresses phenomena. Int J Rock Mech Min Sci 2007;820:834–44.
induce secondary fragmentation due to compaction during the [6] Beaus M, Pariseau W, Stewart B, Iverson S. Design of ore pass. In: Hustrulid W,
gravity flow of the caved rock. The greater the vertical stress Bullock R, editors. Underground mining methods. Litteton, Colorado, USA:
SME; 2001. p. 627–34.
applied, the greater the degree of breakage. The mean size could
[7] Maass S. Technology alternatives to solve hang ups on drawbells [Masters
reduce up to 20% depending on the vertical load applied. thesis]. Santiago: University of Chile; 2013 (in Spanish).
The experimental set up was successful in understanding the [8] Pierce M. A hybrid methodology for secondary fragmentation prediction in
basic mechanisms of flow under confinement. There is more cave mines. In: Potvin Y, editor. Caving 2010. Australia: Perth; 2010. p. 567–81.
[9] Castro R, Trueman R, Halim A. Study of isolated draw zones in block
research to be conducted in the future to further understand the caving mines by means of a large 3D physical model. Int J Rock Mech Min
practical implications of flow under vertical load for mine design Sci 2007;44:860–70.
purposes. Topics to be investigated include the influence of fines [10] Trueman R, Castro R, Halim A. Study of multiple draw-zone interaction in
block caving mines by means of a large 3D physical model. Int J Rock Mech
and of water, as well as the influence of the geometry of drawbells
Min Sci 2008;45:1044–51.
and equipment type on hang up frequency for comparison with [11] Fuenzalida M. Study of the confined gravity flow and its application to caving.
mine scale information. There is also research to be done in terms [Master's thesis]. Santiago, Chile: University of Chile; 2012 ([in Spanish]).
[12] ASTM D7181. Method for consolidated drained triaxial compression test for
of the modeling of breakage and hang ups. Eq. (3) would need
soils. International; 2011.
further refinement to be used as design criteria for caving. In the [13] Janssen H. Experiments regarding grain pressure in silos. In: Karzulovic A,
meantime this type of tests could be considered to be part of a Alfaro M, editors, Proceedings of MassMin. Santiago, Chile; 2004. p. 293–300.